Sie sind auf Seite 1von 7

Course note for EQE 686: Deformation-based seismic assessment and design

Stress Resultant Plasticity Theory


Elastic-Plastic Stiffness Matrix and Plastic Reduction Matrix
for frame element subjected to bending and axial deformation
Prof.Dr. Mehmet Nuray Aydınoğlu
Boğaziçi University, Kandilli Observatory and Earthquake Research Institute
October 2017

A frame element is considered with plastic hinges to be developed at either end or both. For the
sake of simplicity, a two-dimensional formulation is developed, which can be readily extended to
three-dimensional case. In this formulation, material stress – strain relationship is assumed to be
elastic – perfectly plastic, i.e., strain-hardening is neglected.

1. Elastic / plastic displacements and plastic flow rule

When the yield surface at one end of a frame element is reached under incremental loads, a plastic
hinge is formed and the yielding process starts. According to plastic flow rule of plasticity theory,
the response point representing the bending moment and axial force on the yield surface moves
along that surface while yielding continues.

The element total displacement vector increment u(i) at the current incremental step of (i) can
be expressed as:

 u (i)   u (i)
e   up
(i)
(1)

where u e and u (i)


(i)
p represent element elastic and plastic displacement vector increments,

respectively, with components of end rotations ( 1 and 2 ) and end axial displacements
( u1 and u2 ) as shown in Fig.1. 󠄣 Δu2

Δθ2

Δθ1
Δu1
Figure 1. Numbering and sign convention of bending and axial degrees of freedom.

1
Corresponding force components are bending moments ( M 1 and M 2 ) and axial forces ( N1

and N 2 ). Note that shear components at both ends are not shown in Fig.1 and not considered in
the formulation below, as they are related to bending components through equilibrium conditions.

2. Normality criterion

The 󠄣“normality criterion” of plasticity theory states that the plastic displacement increment vector
is normal to the yield surface (Fig.2a,2b). Thus, plastic displacement increment is expressed as

 u (i)
p B h
(i) (i)
(2)

which can be written in partitioned form as


 u p1 
(i)
 B1(i) 0   h1 
(i)

 u   (i)   
(i)
p   (i) 
(3a)

  u p2 
 
 0 B (i)
2 
 
 h2 
(i) (i)
where B k represents the gradient vector and hk is the plastic displacement amplitude for each
element end (k = 1,2). The former is defined as
i 
Sk / M k 
B  (i)
 (k = 1,2) (3b)
 
k
 kS / N k

in which S k represents yield surface at each end (Fig.2a):

Sk (M k , N k )  0 (3c)

As mentioned above, according to plastic flow rule of plasticity theory, the response point
representing M k and N k on the yield surface moves along the surface while yielding continues.
Note that yield surface is always convex. The response point inside the yield surface represents
the elastic case. Response point outside the yield surface is inadmissible.
Sk
hk(i)  f k(i)
N k  u (i) u (i)
ek  u (i)
pk
pk

Sk
hk(i)
M k
Sk (M k , N k )  0

(a) (b) (c)

Figure 2. Yield surface at node k, (a) normality criterion, (b) deformation component vectors, (c) elastic force vector

2
Eq.(3a) can be expressed in explicit form as;

(i)
p1
 S / M 0 
 (i)   1 1
  (i) 
 u p1  up1  S1 / N1
(i)

 u p   (i)    (i)   
(i)
0  h1  (4)
 S 2 / M 2  h2(i) 
 u p2  p2   0

 (i)   0 S / N 2 

up2  2

Note that at the unyielded end of the element, B k  0 will be considered.


(i)

3. Calculation of plastic displacement increment

From Eq.(1), elastic displacement vector increment may be expressed as the difference between
total and plastic displacement increments:

 u (i)
e  u
(i)
  u (i)
p (5)

Thus, element elastic force vector increment  f can be expressed as


 f  k ee  ue (6)

in which k ee represents the element elastic stiffness matrix in which bending and axial degrees of
freedom are uncoupled (Fig.1):

M 1  (i)
e1
  4 EI /L 0 2EI /L 0 
     0
N1  ue1  EA/L 
(i)
EA/L 0
f    ;  u e   (i)  ; k ee  
(i) (i)
(7)
M 2  e2   2 EI /L 0 4EI /L 0 
N 2  u (i)   
 e2   0 EA/L 0 EA/L 

where EI and EA denote bending and axial rigidities and L is the element length. Note that k ee is
constant throughout the analysis and thus independent of incremental step (i). The element elastic
displacement vector  u e and element elastic force vector increment  f are tangent to the yield
surface as shown in Fig.2b, 2c.

From Eqs.(2), (5), (6) element elastic force vector increment can be written as

 f (i)  k ee  u (i)  k ee B (i) h (i) (8)


Applying the normality criterion of plasticity theory;

3
 u (i)T
p f
(i)
0 (9a)

or simply

B (i)T  f (i)  0 (9b)


Substituting Eq.(8) into Eq.(9b);
B (i)T k ee  u (i)  B (i)T k ee B (i) h (i)  0 (10)
Eq.(8) and Eq.(10) can be combined as follows:

 k ee  k ep
(i)
  u (i)   f (i) 
 (i)T   (i)     (11)
  k ep k (i)
pp   
 h 
  0 
where k (i) (i)
ep and k pp are defined as;

k ep  k ee B(i) k pp  B(i)T k ee B(i) (12)

The plastic displacement amplitude vector can be obtained from the second row of Eq.(11) as
1 (i)T
h(i)  k (i)
pp k ep  u
(i)
(13)

and the increment of plastic displacement vector is obtained from Eq.(2) and Eq.(13) as
(i) (i) 1 (i)T
 u (i)
p  B k pp k ep  u
(i)
(14)

4. Derivation of element elastic – plastic stiffness matrix and plastic reduction matrix

After the formation of plastic hinge at one or both ends of an element, its stiffness matrix is to be
modified to represent the effect of hinge formation on the structural system in the forthcoming
increment(s) where new hinge(s) will develop.

(i)
Substituting h given by Eq. (13) into the first row of Eq.(11) results in the elastic-plastic
equilibrium equation at the element level in terms of total displacement increments as:

EP  u
k (i)   f (i)
(i)
(15)
(i)
in which element elastic-plastic stiffness matrix k EP is defined as

EP  k ee  k R
k (i) (i)
(16)
(i)
where the second term k R represents the element plastic reduction matrix, which is defined as

4
(i) (i) 1 (i)T
R  k ep k pp k ep
k (i) (17)

Note that bending and axial degrees of freedom now become coupled in element elastic-plastic
(i)
stiffness matrix k EP in contrast to element elastic stiffness matrix given in Eq.(7).

5. Combining element plastic reduction matrix in global equations of motion for next
incremental step(s)
(i)
The element plastic reduction matrix k R given in Eq.(17) is combined in the system plastic
(i)
reduction matrix K R , which is then substituted into global equations of motion of the structural
system to be used for the next incremental step(s) where plastic hinge(s) is developed elsewhere:

M  uS(i)  C  uS(i)  (K E  K (i)


R  K G )  uS   M I x ugx  M I y ugy  M I z ugz
(i) (i) (i) (i) (i)
(18a)

where M and C represent mass ve viscous damping matrices, respectively, whereas K E , K (i)R
(i)
and K G denote global elastic stiffness matrix, global plastic reduction matrix and global geometric
(i)
stiffness matrix, respectively (Note that plus sign used for axial force-dependent K G is compatible
with axial forces in tension, see Fig.1). uS and time derivatives represent the system
(i)

incremental displacement, velocity and acceleration vectors, respectively. Typical I m (m = x,y,z)


is the rigid transmission vector of ground acceleration component (m), whose elements in m’th 󠄣
direction are unity while others are zero. ugm represents the respective ground acceleration.
(i)

Note that in the case of incremental static analysis (e.g. pushover analysis) Eq.(18a) is reduced to:

(K E  K (i)
R  K G )  uS   P
(i) (i) (i)
(18b)

in which  P
(i)
represents the global incremental static load vector.

6. Practical application with piecewise linear yield surfaces: Incremental Hinge by Hinge
Method

6.1. In practice, yield surfaces are preferably idealized as piecewise linear segments in 2-D and
piecewise linear planes in 3-D response (Ref.[4]). Piecewise linearized 2-D yield surfaces for an
steel wide-flanged I section and a reinforced concrete rectangular section is shown in Fig.3 as
typical examples. Although somewhat approximate, such linearization is acceptable for practice
and enables most of the above-given matrix operations are performed on few number of linearized
segments. Note that normality condition is approximately satisfied in reinforced concrete sections,
yet above formulation is considered reasonably acceptable in practice for incremental static
analysis (e.g. pushover analysis) of reinforced concrete structures.
5
N

Np N

0.15Np
M Balanced
point
Mp M

(a) (b)
Figure 3. Piecewise linearized 2-D yield surfaces of (a) steel I section, (b) Rectangular reinforced concrete section

6.2. After yielding occurred at one end of the frame element, Eq.(8) and Eq.(14) are evaluated to
calculate elastic force and plastic deformation components at the current incremental step. Those
equations as well as plastic reduction matrix (Eq.(17)) of the (i)’th 󠄣increment 󠄣can 󠄣be 󠄣applied 󠄣even 󠄣
in the subsequent increments (in which plastic hinges will develop in other sections), as long as
the response remains on the same piecewise linear segment.

However, if response moves to the next segment of the yield surface (which can be monitored
through current axial force or bending moment), Eq.(14) and Eq.(17) can no longer be used. This
case will be treated as a new increment even if a new hinge is not developed somewhere else, and
all calculations will be renewed for the new increment. Note that such a case occurs rarely in
practice, as long as very short segments are not used in linearization of yield surfaces.

6.3. For static incremental loading, the value of plastic displacement amplitude hk (k = 1,2) is
generally positive. Rarely it turns out to be negative, indicating that unloading occurred due to the
formation of a plastic hinge somewhere, in which case the response turns out to be linear elastic.
In such a case iterations are required to locate the transition point from yielding to elastic response
where hk  0 . Then elastic stiffness coefficients are used for that joint in the stiffness matrix for
the next increment.

In dynamic case unloading always occurs due to cyclic nature of response.

6
References (all for static loading):

[1] McGuire, W., Gallagher, R.H. and Ziemian, R.D., 󠄣 “Matrix Structural Analysis”, 󠄣 Second 󠄣
Edition, Wiley 2000.

[2] Jirásek, M., Bažant, 󠄣Z.P., 󠄣“Inelastic Analysis of Structures”, 󠄣Wiley 󠄣2002.

[3] Wong, 󠄣M.B. 󠄣“Plastic Analysis and Design of Steel Structures”, 󠄣Elsevier 󠄣2009.

[4] Girgin, 󠄣K., 󠄣“A Method of Load Increments for the Determination of Second-Order Limit Load
and Collapse Safety of Reinforced Concrete Framed Structures”, 󠄣Ph.D. 󠄣Thesis, 󠄣Istanbul Technical
University, Faculty of Civil Engineering, 1996 (in Turkish with an extended summary in English).

Das könnte Ihnen auch gefallen